The Statistical Modeling, Causal Inference, and Social Science blog posted an interesting tool (direct link) that attempts to discern your gender identity by evaluating your browser history. Particular websites are weighted based on the known gender ratio of the user base for that particular website and creates a new score. The site assures users it does not store user information and it suggests that there is a 99% chance that I am male versus, obviously a 1% chance that I am female:
Likelihood of you being FEMALE is 1%
Likelihood of you being MALE is 99%
According to this tool, I am 52% female and 48% male. (I wonder if Mike’s 1% comes from me using his computer?)
I think this application reinforces the constraints society places on men and women. I am sure that I am 48% male according to this website because I play online video games and I often search for gaming websites to improve how I play. Does this somehow make me not female? No, but it reinforces society’s idea that I am acting like a man instead of a woman because I enjoy playing video games. Such a silly concept. Women used to be shunned away from playing sports so perhaps in 100 years it will be perfectly acceptable for them to play video games and this application will say they’re 100% female instead of 52%.
It’s not a rating of how female you are, but instead attempts to estimate a probability of you being male or female. It is based purely on the population numbers and seems to use a form of information updating. So, if I only had one piece of information (you visit site A, and site A’s population is 60% female and 40% male) I would naively say that there is a 60% chance you are female. With more information, I can get a better guess.
If you are consistently in the minority, then the guess will be wrong – but that may only be for a minority of cases. For the application to shift, the population visiting the sites has to shift. Also, if you visit sites that is equally visited by men and women (google seems to fall into the category), and do not travel to any extreme population sites – then that will also make it harder for the app to identify you.
That is the point I was trying to make. Since fewer women play video games because society says it is a “guy” thing to do, they will less likely visit the gaming pages I visit and hence the program will indicate I am more male. If society did not dictate the appropriate activities for men and women, more women would visit those pages.
Rather than women being shunned away from playing video games, I thought the stereotype has been video gamers being shunned away from women? (Video games do seem to be rapidly entering the mainstream, however).
According to this program, reading the NY Times is slightly more manly and than playing World of Warcraft and frequenting torrent sites and reading the Motley Fool is about the manliest thing a human could possibly do.
There’s no comment editing on this blog? I hate typos.
i was 61% female and 39% male, so Mike is clearly the alpha in this situation.
If Mike’s the Alpha Male here, then I must be completely the Alpha Female (Tribe Mother? High Priestess? Madonna? What female archetype is appropriate here?).
Likelihood of you being FEMALE is 100%
Likelihood of you being MALE is 0%
Dang. And that’s even with a couple of hits on the highly male-correlated NewEgg.com and TigerDirect.com. I do have Oprah.com, iVillage.com, and Sephora.com in my history, though, so I suppose it washes out.
Some interesting things I noticed: Gmail and NPR both have a fairly gender-mixed online usership, while women make up a greater percentage of visitors to credit reporting websites, BarackObama.com, and CBS.com. Not surprisingly, I think, DrudgeReport.com and Digg.com are more trafficked by men.
I came out as having an 87% probability of being female. The gaming spots I visit don’t have particularly high male-to-female ratios. What did shift the probabilities “against” my actual gender was the fact that I do a lot of our online shopping, and many of the relevant sites skew very heavily female.